This page was translated automatically from Turkish. Read the original

This page is an AI-assisted Turkish summary of the study presented at the 26th World Congress of Psychiatry. The study belongs to the presenting authors; please refer to the original source when citing it.

EV0465

Adaptive Federated Smart Homes with In-Home Displays to Support Mental Well-Being and Sustainable Lifestyles

ADAPTIVE FEDERATED SMART HOMES WITH IN-HOME DISPLAYS FOR MENTAL WELLNESS AND SUSTAINABLE LIFESTYLE ENHANCEMENT

A privacy-preserving smart home framework combining federated learning with in-home display feedback showed signals of improvement in sleep regularity, mental well-being, and lifestyle consistency indicators during a six-week pilot study.

This study, designed to assess feasibility and estimate effect sizes, aimed to investigate whether a privacy-preserving, human-centered smart home could actively support mental well-being rather than merely automate comfort and energy management. The framework combines federated learning, adaptive local algorithms, and in-home display (IHD) feedback to target four behavioral indicators: the Sleep Regularity Index (SRI), Mental Wellness Index (MWI), Lifestyle Consistency Score (LCS), and Stress Event Frequency (SEF). The six-week pilot study was conducted in 15 smart homes, each operating as an independent local learning node equipped with environmental sensors and an IHD interface; data were collected on light exposure, indoor temperature, ambient noise levels, daily activity rhythms, sleep-related patterns, and residents’ self-reported mood or well-being assessments. Raw household data were never transferred to the central server; only learned model parameters were shared for cloud-level federated aggregation, and the improved global model was redistributed to the homes for local personalization. Aggregated pilot results indicated consistent improvements across all target metrics: SRI increased by 26% (more regular sleep timing), MWI increased by 22% (improved perceived mental well-being), LCS increased by 16% (more stable daily routines), and SEF decreased by 18% (fewer stress-related events). The authors note that the sample was deliberately kept small and that the results should therefore be interpreted as signals of feasibility and estimates of effect size rather than definitive clinical evidence; they emphasize the need for longitudinal testing with larger samples, multimodal affect sensing, and more rigorous causal evaluation.

All WCP 2026 summaries